INDEX

Explanations

_DEFAULT

np_max-act · gemini-2.0-flash

references to specific organizations or entities.

oai_token-act-pair · gpt-4o-mini Triggered by @xinyanhu8

This neuron detects the special framing and metadata tokens that structure prompt–completion dialogues (e.g., “<|begin_of_text|>,” “<|start_header_id|>,” user/assistant tags, etc.).

oai_token-act-pair · o4-mini Triggered by @xinyanhu8

New Auto-Interp

Configuration

andyrdt/saes-llama-3.1-8b-instruct/resid_post_layer_11/trainer_1

Dataset (Dashboard)

Various

Features

131,072

Data Type

float32

Hook Name

blocks.11.hook_resid_post

Architecture

standard

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 conseils

-0.07

 Entire

-0.07

 одне

-0.06

tc

-0.06

 conclus

-0.06

.Pin

-0.06

();

↵

-0.06

水

-0.06

addons

-0.06

ptype

-0.06

POSITIVE LOGITS

 jsonData

0.06

ρού

0.06

sæ

0.06

UM

0.06

цю

0.06

 Obesity

0.06

 جمعیت

0.06

 enrol

0.06

gb

0.06

,key

0.06

Activations Density 0.000%

_DEFAULT

references to specific organizations or entities.

This neuron detects the special framing and metadata tokens that structure prompt–completion dialogues (e.g., “<|begin_of_text|>,” “<|start_header_id|>,” user/assistant tags, etc.).

No Comments

No Known Activations

_DEFAULT

references to specific organizations or entities.

This neuron detects the special framing and metadata tokens that structure prompt–completion dialogues (e.g., “<|begin_of_text|>,” “<|start_header_id|>,” user/assistant tags, etc.).

No Comments

No Known Activations